Can we prove that for all natural numbers $n$, there exists a function $\epsilon _{n}$ satisyfing : $\forall t > -1, \sqrt[n]{1+t}=1+\frac{t.\epsilon_{n}(t)}{n}$ and $\lim_{t \to 0}\epsilon _{n}(t)=0$ ?
Tell me more
×
Mathematics Stack Exchange is a question and answer site for
people studying math at any level and professionals in related fields. It's 100% free, no registration required.
|
From the given equation, $$ \epsilon_n = \frac{n}{t}((1+t)^{\frac{1}{n}}-1) \to \frac{(1+t)^{\frac{1}{n}-1}}{1} \to 1$$ as $t \to 0$ by L'Hopital for $"\frac{0}{0}"$. So the answer is No. |
||||
|
|
|
But, by the binomial theorem, for fixed n, $(1+t)^{1/n} = 1 + t/n + (t^2/2)(1/n)(1/n-1) + ... = 1 + t/n + O(t^2)$ so that $\epsilon_n = 1 - t\frac{n-1}{n} + O(t^2) = 1 + O(t)$, not $0$. Your computation $$\epsilon_n = \frac{n}{t}((1+t)^{\frac{1}{n}}-1) \to \frac{(1+t)^{\frac{1}{n}-1}}{1}$$ is correct, but the right-hand limit is $1$ as $t \to 0$, not $0$. |
|||
|
|